Parsing CSV

Walking a line

A CSV line is bytes in a buffer. Parse it with two pointers: one walks the buffer, one marks the start of the current field. The loop ends at the newline or the NUL:

mov rsi, line ; walker mov rdi, line ; field start loop: mov al, [rsi] cmp al, 0 je done

Splitting fields

When the walker hits a comma, the field runs from the field-start pointer to the walker. Copy it, print it, or store its length, then move the field-start pointer past the comma:

cmp al, ',' jne next ; field = [rdi, rsi) lea rdi, [rsi+1] ; next field starts after the comma

Quoted fields

Real CSV quotes fields that contain commas or newlines. A quoted field starts with a double quote, keeps commas as data, and ends at the closing quote. Handle that state machine, or accept only unquoted input and say so in the program's docs.

Next: Data structures: linked list

Article author: Arthur Isaev